作者位次 #3/7ICLR 2026Poster· Deep Learning / Attention Mechanisms
Identifying and Evaluating Inactive Heads in Pretrained LLMs
Pedro Sandoval-Segura · Xijun Wang · Ashwinee Panda · Micah Goldblum · Ronen Basri · Tom Goldstein · David Jacobs
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Pedro Sandoval-Segura · Xijun Wang · Ashwinee Panda · Micah Goldblum · Ronen Basri · Tom Goldstein · David Jacobs
Monte Hoover · Vatsal Baherwani · Neel Jain · Khalid Saifullah · Joseph J Vincent · Chirag Jain · Melissa Rad · C. Bruss · Ashwinee Panda · Tom Goldstein
Nikita Afonin · Nikita Andriianov · Vahagn Hovhannisyan · Nikhil Bageshpura · Kyle Liu · Kevin Zhu · Sunishchal Dev · Ashwinee Panda · Oleg Rogov · Elena Tutubalina · Alexander Panchenko · Mikhail Seleznyov
Recent work has shown that narrow finetuning can produce broadly misaligned LLMs, a phenomenon termed emergent misalignment (EM). While concerning, these findings were limited to finetuning and activation steering, leaving out in-context learning (ICL). We therefore ask: does EM emerge in ICL? We...